Application Report DLPA112–February 2020 DLPC34xx Controller Image Calibration Kyle Rakos, Tony Lee, and Susan Ji ABSTRACT This application report introduces the basics of image calibration when using TI DLP® Technology. First, a brief background describes the concept of white point and what can affect it. Then various approaches are discussed on how it can be changed. It also discusses the calculations needed to adjust white point using duty cycles. Finally, the method to change white point on the DLPC34xx controller in a production environment is detailed. This method may be useful for both design and production engineers and technicians. While this application report is intended for systems using DLPC34xx controllers and associated DMDs (digital micromirror devices), it may be partially useful for any system utilizing DLP technology. Contents 1 Background ................................................................................................................... 2 1.1 White Point .......................................................................................................... 2 1.2 How LED Duty Cycles and Currents Affect White Point....................................................... 2 1.3 Looks ................................................................................................................. 3 1.4 Color Coordinate Adjustment (CCA) ............................................................................. 3 1.5 Gamma............................................................................................................... 3 1.6 Color Science ....................................................................................................... 4 2 Approaches for White Point Correction ................................................................................... 5 2.1 Hardware Domain .................................................................................................. 5 2.2 Software Image Processing Domain ............................................................................. 5 2.3 Light Processing Domain .......................................................................................... 6 3 Calculate the Optimal LED Duty Cycle ................................................................................... 7 3.1 Technique............................................................................................................ 7 3.2 Example.............................................................................................................. 9 4 How to Perform Factory White Point Correction using LED Duty Cycles ............................................ 9 4.1 Factory Tools Required for White Point Calibration .......................................................... 10 4.2 Steps for Adjusting White Point with RGB Duty Cycle Adjustment ......................................... 10 4.3 Example Files...................................................................................................... 11 5 References .................................................................................................................. 12 Trademarks DLP Pico is a trademark of Texas Instruments. DLP is a registered trademark of Texas Instruments. All other trademarks are the property of their respective owners. DLPA112–February 2020 DLPC34xx Controller Image Calibration 1 Submit Documentation Feedback Copyright © 2020, Texas Instruments Incorporated Background www.ti.com 1 Background 1.1 White Point White point, or color temperature, is the color reproduced with full brightness from the red, green, and blue primaries [1]. In other words, it is the color that is considered to be white. For example, older incandescent light bulbs produce a white light that is warmer (more red) compared to newer LED light bulbs which produce a white light that is cooler (more blue). The terms warmer and cooler refer to the use of temperature to determine white point. A typical white point may be 6500K or 5500K. These temperatures correspond to the color emitted by a theoretical black body at the given temperature. These white point temperature values can be plotted on a standard CIE color space graph as shown in Figure 1. Figure 1. Color Space [2] 1.2 How LED Duty Cycles and Currents Affect White Point In DLPC34xx systems, colors are reproduced through time sequential illumination synchronized to the DMD mirrors. A simplified example of displaying an all white image frame is discussed. All DMD mirrors flip into the on position and are illuminated by a red LED, then a green LED, and lastly a blue LED. The human visual system integrates this illumination pattern into a white image. This repeats for each frame. Each illumination period (also referred to as a color's duty cycle) could be divided into equal segments as shown in Figure 2. Assuming a 60-Hz frame rate, each color is shown for (1 / 3) / 60 seconds (5.56 ms). To show a different color, the individual mirrors on the DMD turn off during all, or portions of, an illumination period. For example, yellow can be created by turning on the mirrors for only the red and green illumination time. 2 DLPC34xx Controller Image Calibration DLPA112–February 2020 Submit Documentation Feedback Copyright © 2020, Texas Instruments Incorporated www.ti.com Background Frame Frame 0 Frame 1 Illumination Color Figure 2. Equal Segments Theoretically, an RGB (red, green, blue) duty cycle of 33.33, 33.33, 33.33 can be used (as described above). Often a duty cycle closer to RGB 30, 50, 20 is used in order to achieve the desired white point. In other words, 30% of the time is allocated for red, 50% for green, and 20% for blue as shown in Figure 3. By changing the duty cycles one can therefore change the white point. For example, a higher red duty cycle will shift the white point warmer. Frame Frame 0 Frame 1 Illumination Color Figure 3. Nominal Segments In addition to changing duty cycles, the current of individual LEDs can be varied. As an example, let us assume the nominal LED current is set to 5 A but we desire to create a warmer white point. We could increase the red LED current to 6 A while leaving the green and blue LED current at 5 A (assuming of course 6 A remains within the LED specification). This higher current outputs more red light per unit time and therefore shifts the white point to a warmer level. 1.3 Looks In DLPC34xx systems, various looks can be created which can each have a target white point. A system may have different target white points for various situations (such as a cinema mode and a vivid mode). Each look has a specific duty cycle associated with it that corresponds to the desired white point. Additionally, looks may be further optimized for different situations (such as looks for 2D and looks for 3D). 1.4 Color Coordinate Adjustment (CCA) Digital images are composed of pixels which can be represented by a 24-bit RGB value. Each color is therefore represented by 8-bits and is 0 (full off), 255 (full on), or somewhere in-between. Therefore, to input an all-red pixel to the DLPC34xx system, send a value of (255,0,0). When the DMD recreates this pixel on a DMD mirror, the mirror is on and reflecting light only during the red illumination period. While this behavior can generally be assumed, there are times where it may be desirable to change this behavior. For example, there may be a situation where an input pixel value of RGB (255,0,0) is desired to be transformed into a value of (255,15,0). This enables one to change the white point of the projector through relatively easy firmware settings. However, the CCA is done in nonlinear space and therefore this is not generally a suggested approach. 1.5 Gamma As previously mentioned, digital images are composed of pixels which can be represented by a 24-bit RGB value. Each color is therefore either 0 (full off), 255 (full on), or somewhere in-between. While one may expect a pixel value of 128 to correspond to a human eye perceived brightness of 50% and a light energy output of 50%, in reality the human eye perceived brightness and light energy output are not the same. Gamma encoding takes into account this non-linear relationship between light energy and human eye perceived brightness. DLPA112–February 2020 DLPC34xx Controller Image Calibration 3 Submit Documentation Feedback Copyright © 2020, Texas Instruments Incorporated Background www.ti.com A pixel value of RGB (128,128,128) that uses standard gamma-encoding will appears to the human eye as 50% of full brightness. However, before being displayed, a gamma decoding function (sometimes called a degamma function) is used to change the light energy output to 22% of maximum brightness. Therefore, the DLPC34xx controller takes the gamma encoded value and applies a gamma decoding function to the value before display on the DMD. The below equation takes a gamma encoded decimal value (Vencoded) between 0 and 1 and outputs a decoded decimal value (Vdisplay) for display between 0 and 1. This can be scaled for any input and output value range (such as 0 to 255) [3]. @ Vdisplay = Vencoded A typical degamma curve has γ = 2.2 as shown in Figure 4. Different values of γ can be used for different gamma encoding schemes. Additionally, various degamma curves can be created for specialized use cases and system optimization. 255 210 180 150 120 90 60 Output (Displayed)PixelValue 30 0 0 30 60 90 120 150 180 210 255 Input (Gamma Encoded) Pixel Value Figure 4. Typical Degamma Curve (γ = 2.2) 1.6 Color Science Colors are often measured